by Paul E. McKenney

[permalink] [raw]

Subject: Re: [PATCH RFC tip/core/rcu] Fix attempt to avoid offloading callbacks unless requested

On Mon, Jul 28, 2014 at 07:35:55PM -0400, Pranith Kumar wrote:
> On Mon, Jul 28, 2014 at 7:17 PM, Paul E. McKenney
> <[email protected]> wrote:
> > On Mon, Jul 28, 2014 at 11:03:06PM +0200, Frederic Weisbecker wrote:
> >>
> >> Paul what do you think? If we keep that behaviour, Maybe you could blindly do
> >> rcu_nocb_mask |= tick_nohz_full and remove the CONFIG_NO_HZ_FULL_ALL dependency
> >> on RCU_NOCB_ALL?
> >
> > You mean something like the following updated patch?
> >
> > Thanx, Paul
> >
> > ------------------------------------------------------------------------
> >
> > rcu: Fix attempt to avoid unsolicited offloading of callbacks
> >
> > Commit b58cc46c5f6b (rcu: Don't offload callbacks unless specifically
> > requested) failed to adjust the callback lists of the CPUs that are
> > known to be no-CBs CPUs only because they are also nohz_full= CPUs.
> > This failure can result in callbacks that are posted during early boot
> > getting stranded on nxtlist for CPUs whose no-CBs property becomes
> > apparent late, and there can also be spurious warnings about offline
> > CPUs posting callbacks.
> >
> > This commit fixes these problems by adding an early-boot rcu_init_nohz()
> > that properly initializes the no-CBs CPUs.
> >
> > Note that kernels built with CONFIG_RCU_NOCB_CPU_ALL=y or with
> > CONFIG_RCU_NOCB_CPU=n do not exhibit this bug. Neither do kernels
> > booted without the nohz_full= boot parameter.
> >
> > Signed-off-by: Paul E. McKenney <[email protected]>
> > Reviewed-by: Pranith Kumar <[email protected]>
> >
> > diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
> > index d231aa17b1d7..cc7bed1c90dc 100644
> > --- a/include/linux/rcupdate.h
> > +++ b/include/linux/rcupdate.h
> > @@ -269,6 +269,14 @@ static inline void rcu_user_hooks_switch(struct task_struct *prev,
> > struct task_struct *next) { }
> > #endif /* CONFIG_RCU_USER_QS */
> >
> > +#ifdef CONFIG_RCU_NOCB_CPU
> > +void rcu_init_nohz(void);
> > +#else /* #ifdef CONFIG_RCU_NOCB_CPU */
> > +static inline void rcu_init_nohz(void)
> > +{
> > +}
> > +#endif /* #else #ifdef CONFIG_RCU_NOCB_CPU */
> > +
> > /**
> > * RCU_NONIDLE - Indicate idle-loop code that needs RCU readers
> > * @a: Code that RCU needs to pay attention to.
> > diff --git a/init/Kconfig b/init/Kconfig
> > index 41066e49e880..6944acd00dff 100644
> > --- a/init/Kconfig
> > +++ b/init/Kconfig
> > @@ -737,7 +737,7 @@ choice
> >
> > config RCU_NOCB_CPU_NONE
> > bool "No build_forced no-CBs CPUs"
> > - depends on RCU_NOCB_CPU && !NO_HZ_FULL_ALL
> > + depends on RCU_NOCB_CPU
> > help
> > This option does not force any of the CPUs to be no-CBs CPUs.
> > Only CPUs designated by the rcu_nocbs= boot parameter will be
> > @@ -751,7 +751,7 @@ config RCU_NOCB_CPU_NONE
> >
> > config RCU_NOCB_CPU_ZERO
> > bool "CPU 0 is a build_forced no-CBs CPU"
> > - depends on RCU_NOCB_CPU && !NO_HZ_FULL_ALL
> > + depends on RCU_NOCB_CPU
> > help
> > This option forces CPU 0 to be a no-CBs CPU, so that its RCU
> > callbacks are invoked by a per-CPU kthread whose name begins
> > diff --git a/init/main.c b/init/main.c
> > index e8ae1fef0908..5d8c83ae6c55 100644
> > --- a/init/main.c
> > +++ b/init/main.c
> > @@ -577,6 +577,7 @@ asmlinkage __visible void __init start_kernel(void)
> > idr_init_cache();
> > rcu_init();
> > tick_nohz_init();
> > + rcu_init_nohz();
> > context_tracking_init();
> > radix_tree_init();
> > /* init some links before init_ISA_irqs() */
> > diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h
> > index 00dc411e9676..91208497e407 100644
> > --- a/kernel/rcu/tree_plugin.h
> > +++ b/kernel/rcu/tree_plugin.h
> > @@ -85,33 +85,6 @@ static void __init rcu_bootup_announce_oddness(void)
> > pr_info("\tBoot-time adjustment of leaf fanout to %d.\n", rcu_fanout_leaf);
> > if (nr_cpu_ids != NR_CPUS)
> > pr_info("\tRCU restricting CPUs from NR_CPUS=%d to nr_cpu_ids=%d.\n", NR_CPUS, nr_cpu_ids);
> > -#ifdef CONFIG_RCU_NOCB_CPU
> > -#ifndef CONFIG_RCU_NOCB_CPU_NONE
> > - if (!have_rcu_nocb_mask) {
> > - zalloc_cpumask_var(&rcu_nocb_mask, GFP_KERNEL);
> > - have_rcu_nocb_mask = true;
> > - }
> > -#ifdef CONFIG_RCU_NOCB_CPU_ZERO
> > - pr_info("\tOffload RCU callbacks from CPU 0\n");
> > - cpumask_set_cpu(0, rcu_nocb_mask);
> > -#endif /* #ifdef CONFIG_RCU_NOCB_CPU_ZERO */
> > -#ifdef CONFIG_RCU_NOCB_CPU_ALL
> > - pr_info("\tOffload RCU callbacks from all CPUs\n");
> > - cpumask_copy(rcu_nocb_mask, cpu_possible_mask);
> > -#endif /* #ifdef CONFIG_RCU_NOCB_CPU_ALL */
> > -#endif /* #ifndef CONFIG_RCU_NOCB_CPU_NONE */
> > - if (have_rcu_nocb_mask) {
> > - if (!cpumask_subset(rcu_nocb_mask, cpu_possible_mask)) {
> > - pr_info("\tNote: kernel parameter 'rcu_nocbs=' contains nonexistent CPUs.\n");
> > - cpumask_and(rcu_nocb_mask, cpu_possible_mask,
> > - rcu_nocb_mask);
> > - }
> > - cpulist_scnprintf(nocb_buf, sizeof(nocb_buf), rcu_nocb_mask);
> > - pr_info("\tOffload RCU callbacks from CPUs: %s.\n", nocb_buf);
> > - if (rcu_nocb_poll)
> > - pr_info("\tPoll for callbacks from no-CBs CPUs.\n");
> > - }
> > -#endif /* #ifdef CONFIG_RCU_NOCB_CPU */
> > }
> >
> > #ifdef CONFIG_TREE_PREEMPT_RCU
> > @@ -2451,6 +2424,67 @@ static void do_nocb_deferred_wakeup(struct rcu_data *rdp)
> > trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu, TPS("DeferredWakeEmpty"));
> > }
> >
> > +void __init rcu_init_nohz(void)
> > +{
> > + int cpu;
> > + bool need_rcu_nocb_mask = true;
> > + struct rcu_state *rsp;
> > +
> > +#ifdef CONFIG_RCU_NOCB_CPU_NONE
> > + need_rcu_nocb_mask = false;
> > +#endif /* #ifndef CONFIG_RCU_NOCB_CPU_NONE */
> > +
> > +#if defined(CONFIG_NO_HZ_FULL) && !defined(CONFIG_NO_HZ_FULL_ALL)
> > + if (tick_nohz_full_running && cpumask_weight(tick_nohz_full_mask))
> > + need_rcu_nocb_mask = true;
> > +#endif /* #if defined(CONFIG_NO_HZ_FULL) && !defined(CONFIG_NO_HZ_FULL_ALL) */
>
> If you are removing the dependency on NO_HZ_FULL_ALL, I think this can just be
>
> #ifdef CONFIG_NO_HZ_FULL
>
> Otherwise, a kernel config with CONFIG_RCU_NOCB_NONE=y,
> CONFIG_NO_HZ_FULL_ALL=y with nohz_full= boot parameter will not have a
> rcu_nocb_mask.

Good point, fixed.

Thanx, Paul